Cohort Analysis for SaaS Founders: Read Retention
Published on March 13, 2026 · Jules, Founder of NoNoiseMetrics · 9min read
Your March cohort churns at 40% in 90 days. Your June cohort churns at 12%. Same product. Same price. Something changed.
Cohort analysis is the tool that shows you this. It groups customers by when they signed up and tracks their behavior over time — so you can see which versions of your product retain customers and which don’t.
This guide explains how it works, how to build one, and how to read it. For context on why churn matters in the first place, read the complete guide to SaaS churn.
Table of Contents
- What Is Cohort Analysis and Why It Matters
- What Does a Cohort Table Look Like?
- How to Build a Cohort Table from Stripe
- How to Read and Interpret a Cohort Table
- Revenue Cohort vs Customer Cohort
- When Cohort Analysis Changes Your Decisions
- Advanced Cohort Analysis
- FAQ
What Is Cohort Analysis and Why It Matters
Cohort analysis groups customers by a shared characteristic — usually their acquisition date — and tracks a specific behavior (like retention or revenue) over time for each group.
- A cohort = customers who signed up in the same month (or week, or product version)
- The analysis = what % of each cohort is still active at 30, 60, 90, 180 days
Why it beats aggregate metrics: aggregate retention hides the fact that your new cohorts may be terrible while old ones hold the average up. A 75% retention rate could mean every cohort retains at 75% — or it could mean old cohorts retain at 90% while new ones retain at 60%.
You can’t fix what you can’t see. Use cohort analysis to calculate your overall retention rate in context.
What Does a Cohort Table Look Like?
| Cohort | Month 0 | Month 1 | Month 2 | Month 3 |
|---|---|---|---|---|
| Jan 2025 | 100% | 72% | 61% | 55% |
| Feb 2025 | 100% | 68% | 57% | 51% |
| Mar 2025 | 100% | 81% | 74% | 70% |
Each row = a signup cohort. Each column = how many are still paying N months later.
The March cohort is visibly stronger — something changed for the better in March. Maybe you launched a new onboarding flow. Maybe a product bug was fixed. Maybe a new blog post attracted better-fit customers. The table surfaces the change. You have to diagnose the cause.
This is what cohort retention means in practice: not an aggregate number, but a table that shows the health of each batch of customers over time.
How to Build a Cohort Table from Stripe
Step-by-step (minimal version):
- Export customers with their
createddate from Stripe - Export all
invoice.paidevents per customer - For each customer: mark which months they paid (month 0 = signup month, month 1 = one month later, etc.)
- Group customers by their signup month (cohort)
- Calculate % of each cohort that paid in each subsequent month
Honest note: this is tedious to do manually. Stripe doesn’t build this view natively.
Tools that do this automatically: NoNoiseMetrics, ChartMogul, Baremetrics. For a cohort analysis template for Stripe data, start with a basic dashboard setup first.
How to Read and Interpret a Cohort Table
How to interpret cohort analysis: look for these four patterns. This approach aligns with David Skok’s SaaS metrics framework — retention patterns tell you more than any single number.
The Retention Cliff (bad)
Rapid drop in Month 1 (e.g., 100% → 40%) = activation problem. Customers signed up but never saw value. They’re not cancelling because they’re unhappy — they’re cancelling because they never actually started.
Fix: onboarding sequence, activation milestone.
The Slow Leak (bad but fixable)
Gradual decline: 100% → 85% → 72% → 61%. Steady churn with no floor. Likely a product-market fit issue or competing alternatives eroding your position.
Fix: exit surveys, feature gap analysis. But this takes longer — you’re fighting product drift.
The Smile Curve (good)
Steep early drop, then flattens: 100% → 65% → 60% → 59%. Customers who survive Month 1 stay essentially forever. This is the pattern you want. Your churn problem is an activation problem, not a retention problem — which is much easier to fix.
Fix: improve Month 1 activation. Everything downstream improves automatically.
Improving Cohorts (great)
Each newer cohort retains better than the previous. March is better than February, which is better than January. Something you changed is working.
This is the signal to run toward — find the change and double down on it.
Revenue Cohort Analysis vs Customer Cohort Analysis
Same concept, different measurement unit:
| Metric | What It Tracks | When to Use |
|---|---|---|
| Customer cohort | % of customers still active | Uniform pricing |
| Revenue cohort | % of starting MRR still being paid | Mixed pricing plans |
Revenue cohort analysis is more useful if you have multiple price tiers. It shows whether your highest-value customers churn faster or slower than average — which is a completely different diagnostic question.
Revenue Cohort Table: A Worked Example
Here’s what a revenue cohort table looks like in practice:
| Cohort | Starting MRR | Month 1 | Month 2 | Month 3 | Month 4 | Month 5 | Month 6 |
|---|---|---|---|---|---|---|---|
| Oct 2025 | €4,200 | 91% | 85% | 82% | 80% | 79% | 78% |
| Nov 2025 | €3,800 | 88% | 80% | 74% | 71% | — | — |
| Dec 2025 | €5,100 | 93% | 89% | 86% | — | — | — |
| Jan 2026 | €4,600 | 90% | 84% | — | — | — | — |
| Feb 2026 | €3,200 | 87% | — | — | — | — | — |
| Mar 2026 | €4,900 | 94% | — | — | — | — | — |
Notice that the December and March cohorts retain noticeably better than the others. That’s your signal to investigate: what changed? Did you adjust pricing? Ship a new onboarding flow? Shift acquisition channels? The table surfaces the anomaly — your job is to find the cause and replicate it.
Revenue cohorts reveal something customer cohorts hide: whether high-value customers churn faster or slower than low-value ones. If your €99/mo customers leave faster than your €19/mo customers, you have a value-fit problem at the top of your pricing. The product delivers enough value at €19 but not enough to justify €99. That’s a different fix than “reduce churn” — it’s “fix the value proposition for your best plan.”
For a deeper look at how revenue retention affects your overall business health, see NRR: Net Revenue Retention for Bootstrappers.
When Cohort Analysis Changes Your Decisions
Numbers are interesting. Decisions are useful. Here’s when cohort data actually changes what you do.
1. Validating an onboarding experiment
You redesigned onboarding in February. Your March cohort shows Month-1 retention at 81%, up from 72% in the January and February cohorts. That’s not noise — that’s a 9-point improvement in the most critical retention window. Don’t second-guess it. Don’t wait for “more data.” Double down on the new onboarding and keep refining it.
2. Measuring channel quality
Your organic cohorts retain at 85% at Month 3. Your paid cohorts retain at 55% at Month 3. Same product, same pricing, same onboarding — completely different retention outcomes. The difference is audience quality. Paid ads brought people who were curious but not committed. Organic brought people who were already searching for a solution.
This doesn’t mean “stop all paid.” It means either fix the targeting (narrower audience, better qualifying copy) or accept that paid acquisition has a built-in retention penalty and price your CAC accordingly.
3. Reading the impact of a price change
You raised prices in January. If the January cohort retains at the same rate or better than the December cohort, the higher price selected for better-fit customers — people who saw enough value to pay more. If January retention drops, the price increase attracted more price-sensitive customers who were less committed from the start.
Cohort analysis turns a pricing change from a guess into a measured experiment.
The difference between aggregate retention and cohort analysis is the difference between revenue churn and customer churn — one hides the story, the other tells it.
Advanced Cohort Analysis — Going Deeper
Once you can read a basic cohort table, segment it. As a16z’s 16 SaaS Metrics highlights, the best operators go beyond top-line numbers and drill into cohort-level behavior:
- By acquisition channel: organic vs paid vs referral. Which channel brings customers who stay?
- By pricing plan: monthly vs annual. Annual plans almost always show better retention.
- By company size (if B2B): do SMBs churn faster than mid-market customers?
- By customer segment: group customers by behavior, plan, or use case to surface which segments retain best.
Goal: find the cohort segment with the best retention, then figure out what’s different about those customers. That’s your ideal acquisition target.
FAQ
What is cohort analysis used for in SaaS?
In SaaS, cohort analysis is used to measure customer retention over time for groups of customers who signed up in the same period. It reveals whether new product improvements are actually improving retention, and which acquisition channels bring the most loyal customers.
How often should I run cohort analysis?
Monthly is the standard cadence for most bootstrapped SaaS products. If you’re making significant product changes, run it weekly to see the impact on recent cohorts faster.
What’s the difference between cohort analysis and retention rate?
Retention rate is a single aggregate number. Cohort analysis is a table that shows retention for multiple groups over multiple time periods. Cohort analysis is what lets you see which customers are driving your retention rate up or down.
Do I need a data team to do cohort analysis?
No. A spreadsheet can do basic cohort analysis if you can export Stripe data. Tools like NoNoiseMetrics or ChartMogul build it automatically. The math is simple — it’s the data extraction that’s tedious manually.
Can I do cohort analysis in a spreadsheet?
Yes, but it’s tedious. Export Stripe customers by signup month, match against paid invoices, and calculate retention manually. A spreadsheet handles 50 customers fine. At 200+, you’ll want automation — cross-referencing invoices with signup dates across multiple months becomes error-prone fast. Tools like NoNoiseMetrics build cohort tables from Stripe data automatically, updated every sync.
What is a good Month-1 retention rate?
For B2B SaaS, Month-1 retention above 80% is strong. Between 65-80% is workable but indicates an onboarding gap — customers signed up with intent but didn’t reach the activation moment fast enough. Below 65% means most customers never experienced enough value to justify paying again. That’s an activation problem, not a retention problem — and the fix is in your first-run experience, not your product roadmap.
How do I use cohort analysis with NRR?
Cohort analysis shows you which groups retain. NRR tells you whether your existing customer base grows or shrinks in revenue terms. Together, they answer: are newer cohorts retaining well enough to sustain the growth rate? If NRR is below 100% and cohort retention is declining month over month, you have a compounding problem — each new batch of customers is weaker than the last, and your existing base is shrinking. See the NRR guide for bootstrappers for the full picture.
What is cohort analysis?
Cohort analysis groups customers by their signup date (or another shared trait) and tracks their behavior over time. It reveals whether newer cohorts retain better than older ones, helping you measure the real impact of product changes on long-term retention. Unlike aggregate retention rates, cohort analysis shows the trajectory of each batch of customers separately — so you can see whether your product is getting better at keeping people, or worse.
Your cohort table is hiding in your Stripe data. NoNoiseMetrics builds it automatically — see which cohorts leak and which ones stick. Free up to €10k MRR →
Free Tool
Try the MRR Dashboard Template →
Interactive calculator — no signup required.
Sources: OpenView 2024 SaaS Benchmarks, Bessemer State of the Cloud 2024, David Skok — SaaS Metrics 2.0, a16z — 16 Startup Metrics